International Journal of Computer and Electrical Engineering, Vol. 6, No.

4, August 2014

A Novel Watermarking Algorithm for Color Images Based

on Discrete Wavelet Transform
Anubhav Kumar, Member, IACSIT, and Anuradha

AbstractIt is very easy to copy and distributed the digital

data illegally. There are many methods for solving this issue like
intellectual, material right protection and Copy right protection
for authors, owners, distributors. As the advancement, invisible
double digital watermarking technique has received
considerable attention. This paper gives a robust digital
watermarking algorithm for color images based on discrete
wavelet transform. Here, we choose a suitable method to
watermarking algorithm for color images based on Haar
discrete wavelet transform. The experiment result are very well,
clearly represent the transparence and robustness of the
algorithm. The experiment result shown that this algorithm is
also secure to various attack.
Index TermsWatermarking, Haar wavelet, DWT, IDWT,
PSNR, NC.

I. INTRODUCTION
Digital watermarking technology is basically known for the
practical importance in the intellectual property rights. The
forthcoming technology has become more advanced and fast
development by the use of digital watermarking. In the recent
years the digital watermarking has achieved result in the field
of correlation-based techniques, frequency domain
techniques, DCT based techniques, DFT based techniques
and DWT based techniques. Digital watermarking is a
technique of embedding some information into the given data,
which is extracted later or being detected for various purposes.
The data embedded is called as watermark and the given data
is original information. Song Qiang and Zhang Hongbin [1]
proposed a new color image watermarking scheme based on
image self-embedding techniques. Song Qiang and Zhang
Hongbin techniques based on the bilinear interpolation and
DWT). In the recent tool in the field watermarking is Singular
value decomposition. Chih-Chin Lai and Cheng-Chih Tsai [2]
are proposed a hybrid image watermarking scheme based on
discrete wavelet transform (DWT) and singular value
decomposition
(SVDTwo-Level
DCT
and
SVD
watermarking for copyright protection and authenticity is
proposed by Jian Yu Bai and Feng Liu [3]. Most of the
domain transformation watermarking schemes works with
DWT and DCT [4].
Digital watermarking scheme for color image based on
image watermarking scheme for color image based on image
Manuscript received January 5, 2014; revised March 8, 2014.
Anubhav Kumar is with the Raj Kumar Goel Institute of Technology for
Women, Ghaziabad, India (e-mail: rajput.anubhav@gmail.com).
Anuradha is with the Electronics and Communication Engineering
Department, Laxmi Devi Institute of Engineering and Technology, Alwar,
India (e-mail: anuradhamit@gmail.com).

DOI: 10.7763/IJCEE.2014.V6.843

303

fusion using wavelet transform based on R, G, B color space

is proposed by Zhou Zude et al. [5]. A novel DMWT
algorithm of DCIW is proposed by Shujiao Liao, which focus
on the original RGB color space and used it.
G. Sun and Y. Yu [6] proposed a digital watermarking
embedding algorithm based on discrete wavelet transform
(DWT) for color image. The original color image is
transformed into the YIQ color space, and the watermark was
embedded in both Y component and Q component. P. S.
Huang et al. [7] presents an approach for hiding the
watermark into DC components of the colour image directly
in the spatial domain, followed by a saturation adjustment
technique performed in RGB space.
D. Kundur and D. Hatzinakos [8] presents a novel robust
watermarking approach called FuseMark based on the
principles of image fusion for copy protection or robust
tagging applications and C. S. Lu and H. Y. M. Liao [9]
proposed a novel multipurpose watermarking scheme, in
which robust and fragile watermarks are simultaneously
embedded, for copyright protection and content
authentication. Watermarking using discrete multiwavelet
transform proposed, where L. Ghouti et al. [10] presents a
robust watermarking algorithm using balanced multiwavelet
transform and P. Kumswat et al. [11] proposed the spread
spectrum image watermarking algorithm using the discrete
multiwavelet transform.
The operation course of our technique only involves Haar
discrete wavelet transform and Inverse Haar discrete wavelet
transform to the R, G, B elements of color original image and
watermark image respectively. The reminder of this paper is
organized as follows: In Section II, the proposed scheme,
including the watermark generation, embedding, and
extraction is introduced. In Section III, the experimental
results and analysis are presented to show the effectiveness of
this scheme. Conclusions are finally drawn in Section IV.

II. PROPOSED ALGORITHM

A. Watermark Embedding Algorithm
The procedure of the watermarking generation and
embedding process is described in Fig. 1. Step for embedding
algorithm steps are as follows:
Step 1: In our proposed algorithm first of all watermark
image and the original color image both are resolved to three
colors RGB channels.
Step 2: After first step, DWT is taken for each layers of
original and watermark image. Further two levels Haar
discrete wavelets transform is used. In Fig. 2, the indexing of
subbands in two level wavelet decomposition is described.

International Journal of Computer and Electrical Engineering, Vol. 6, No. 4, August 2014

Fig. 1. Watermark embedding algorithm block diagram.

Fig. 3. Watermark extraction algorithm block diagram.

Fig. 2. Two-level wavelet decomposition.

Step 3: After discrete wavelet transform, the watermark

information is embedded in to the original image and a two
dimensional image is obtained. A good visual quality is
achieved using a different coefficient for embedding the
watermark information. This algorithm shows that the
original color digital image is decomposed into three color
components of O(r), O(g), O(b), respectively, resulting in
three color components of two layer of wavelet
decomposition.
Step 4: The decomposed color components obtain are W(r),
W(g), W(b) respectively for same color digital watermark.
The decomposition coefficients corresponding to this original
color image obtained by embedding the digital watermark of
three primary color (RGB) used are in the formula given
below [12]:
O(r) A(i, j) = O(r) A(i, j)+1w(r) A(i, j)

(1)

O(g) A(i, j) = O(g) A(i, j)+2w(g) A(i, j)

(2)

O(b) A(i, j) = O(b) A(i, j)+3w(b) A(i, j)

(3)

The parameter 1, 2, 3 are called R, G, B embedding

intensity. Where A (i, j) is the original image of the red, green,
blue, component. These affect the validity of the algorithm
directly.
Step 5: Reconstruction the processed image use the inverse
discrete wavelet transform (IDWT) for dealing with the
watermarked image in R, G, B separate channel,
Step 6: Color images are combined into R, G, and B
through channel combination approach. So the final
watermarked image is obtained. According to the human eye
sensitivity varies through different colors by R: G: B = 2:1:4.

Step 1: Firstly in watermark extraction algorithm, the

channel separation operation is applied on the watermarked
color image and original image to generate its sub images.
Step 2: For the approximate coefficients and detail
coefficients the 2-level discrete wavelet transform is applied
on the sub images.
Step 3: After a DWT transform for the two-dimensional
image, the watermark information is extracting from the
watermarked image. According to the algorithm original
color digital image is decomposed into three color
components of O(r), O(g), O(b), respectively.
Step 4: Finally original watermarked image decomposes
into Wd(r), Wd(g), Wd(b) are three color components,
respectively. In this process formula for the extraction
algorithm is [12]:
O(r) A(i, j) = O(r) A(i, j)+ (wd(r) A(i,j)) /1

(4)

O(g) A(i, j) = O(g) A(i, j)+ (wd(g) A(i,j))/ 2

(5)

O(b) A(i,j) = O(b) A(i, j)+ (wd(b) A(i,j))/ 3

(6)

Step 5: Three watermark images (RGB) are extracted after

the execution of above algorithm and application of inverse
two levels Haar discrete wavelet transform.
Step 6: The original watermark image is obtained by
combining the three R, G, and B watermark image.

III. EXPERIMENT AND RESULTS

The experiment is done with a PC having configuration of
2.4 GHz personal laptop having 2GB RAM and Matlab 7.0
software platform is use to perform the experiment. The test
set has been selected randomly for internet evaluation
experiment watermark image. In the proposed algorithm
experiment, we take 512512 Lena and Girl as an example.
The watermark is a visually recognizable color image which is
256256 Ludong University logo. Fig. 4(a) and Fig. 5(a)
shows the Original image and Fig. 4(b) and Fig. 5(b) shows
Watermark image. Fig. 4(c) and Fig. 5(c) shows the
watermarked image that is derived by embedding the
watermark into the original image. Comparing Fig. 4(a), Fig.
5(a) and Fig. 4(c), Fig. 5(c), we can see that these two images
look almost the same. Fig. 4(d) and Fig. 5(d) show the
recovered watermark image from proposed methods.

B. Watermark Extraction algorithm

The extraction algorithm is the inverse of the embedding
process shown in Fig. 3. Watermark is the seed value for
authorized users at the receiver. The Watermark Extraction
algorithm steps are as follows:
304

International Journal of Computer and Electrical Engineering, Vol. 6, No. 4, August 2014

(a)

(b)

(e)

(f)

(g)

(h)

(i)

(j)

(c)
(d)
Fig. 4. (a) Original girls image, (b) watermark image, (c) watermarked
image, (d) recovered image.

(a)

(b)

(k)

(l)

(m)

(n)

(c)
(d)
Fig. 5. (a) Original Lena image, (b) watermark image, (c) watermarked
image, (d) recovered image.

In experiment of proposed algorithm for test the robustness,

we experiments with the watermarked image by using
common image processing operations such as adding salt and
pepper noise, Gaussian noise, Brightness and contrast
increment, Rotation on 45and 90and so on. The extracted
images in Fig. 6 show the experimental results with these
attacks. Experiment results show that the algorithm has robust
when faced with attacks like salt and pepper noise, Gaussian
noise, Brightness and contrast increment, Rotation.

(a)

(o)
(p)
Fig. 6. (a) Pepper and salt noise(.01) watermarked, (b) recovered pepper and
salt noise (.01) watermark, (c) pepper and salt noise(.02) watermarked, (d)
recovered pepper and salt noise (.02) watermark, (e) Gaussian noise (.001)
watermarked, (f) recovered Gaussian noise (.001) watermark, (g) Gaussian
noise (.002) watermarked, (h) recovered Gaussian noise (.002) watermark, (i)
rotation 45watermarked, (j) recovered rotation 45watermark, (k) rotation
90 watermarked, (j) recovered rotation 90 watermark, (m) brightness
(0.03,1) watermarked, (n) recovered brightness (0.03,1) watermark, (o)
contrast (0.06,0.98) watermarked, (p) recovered contrast (0.06,0.98)
watermark.

(b)

In this paper, the normalized correlation NC between

extracted watermark image (W ) and original watermark (W)
is used to evaluate the watermarking algorithm.
i 1
M

NC
(c)

(d)

305

N
j 1

M
i 1

W ( i , j )W ( i , j )

M
j 1

W (i , j )

(7)

International Journal of Computer and Electrical Engineering, Vol. 6, No. 4, August 2014

Compute the peak-signal-to-noise ratio (PSNR) from 8,

between the original image X and the watermarked image Y.

and multimedia.
REFERENCES

P S N R 1 0 lo g (

(M N )

S. Chen, B. Mulgrew, and P. M. Grant, A clustering technique for

digital communications channel equalization using radial basis
function networks, IEEE Trans. on Neural Networks, vol. 4, pp.
570-578, July 1993.
[2] Q. Song and H. B. Zhang, Color image self-embedding and
watermarking based on DWT, in Proc. International Conference on
Measuring Technology and Mechatronics Automation, 2010, vol. 1,
pp. 796 799.
[3] C. C. Lai and C. C. Tsai, Digital image watermarking using discrete
wavelet transform and singular value decomposition, IEEE
Transactions on Instrumentation and Measurement, vol. 59, no. 11, pp.
30603063, 2010.
[4] J. Y. Bai and F. Liu, A novel grayscale watermarking algorithm, in
Proc. 2nd International Conference on Information Engineering and
Computer Science, 2010, pp. 1-4.
[5] W. C. Chu, DCT based image watermarking using sub sampling,
IEEE Trans. on Multimedia, vol. 5, pp. 34-38, 2003.
[6] G. Sun and Y. Yu, DWT based watermarking algorithm of color
images, in Proc. 2nd IEEE Conference on Industrial Electronics and
Applications, May 2007, pp. 1823-1826.
[7] P. S. Huang, C. S. Chiang, C. P. Chang, and T. M. Tu, Robust spatial
watermarking technique for colour images via direct saturation
adjustment, Vision, Image and Signal Processing, vol. 152, issue 5,
pp. 561-574, 2005.
[8] D. Kundur and D. Hatzinakos, Towards robust logo watermarking
using multiresolution image fusion, IEEE Transcations on
Multimedia, vol. 6, issue 1, pp. 185-197, 2004.
[9] C. S. Lu and H. Y. M. Liao, Multipurpose watermarking for image
authentication and protection, IEEE Transaction on Image
Processing, vol. 10, pp. 1579-1592, Oct. 2001.
[10] L. Ghouti, A. Bouridane, M. K. Ibrahim, and S. Boussakta, Digital
image watermarking using balanced multiwavelets, IEEE Trans.
Signal Process., vol. 54, no. 4, pp. 1519-1536, 2006.
[11] P. Kumswat, K. Attakitmongcol, and A. Striaew, A new approach for
optimization in image watermarking by using genetic algorithms,
IEEE Transactions on Signal Processing, vol. 53, no. 12, pp.
4707-4719, December 2005.
[12] Z. Zhou, Q. S. Ai, and Q. Liu, Digital watermarking scheme for color
image based on image fusion, in Proc. IET International Conference
on Wireless, Mobile and Multimedia Networks, 2006, pp. 13.
[1]

(8)

M SE

where
M SE

1
MN

i 1 i 1 (Y ( i , j ) X ( i , j ))
M

(9)

The watermark extracted can be after identified even after

attacks using results of Table I and Table II. This proves the
robustness of the algorithm against of noisy, geometric and
physical attacks.
TABLE I: EXPERIMENT RESULTS AND ANALYSIS ON GIRLS IMAGE
Attack types
Parameters
Watermarked
NC
PSNR/db
Proposed
36.691
0.9982
Original
0.01
29.69
0.9312
pepper-and-sal
t noise
0.02
27.22
0.8714
0.001
33.415
0.9344
Gaussian noise
0.002
31.610
0.8869
45
38.410
0.6993
Rotation
90
36.690
0.8801
(0.03,1)
32.580
0.8662
Brightness
(.06,0.98)
35.291
0.9733
Contrast
TABLE II: EXPERIMENT RESULTS AND ANALYSIS ON LENA IMAGE
Attack types
Parameters
Watermarked NC
PSNR/db
Proposed
36.699
0.9897
Original
0.01
30.176
0.9154
pepper-and-sal
t noise
0.02
27.665
0.8502
0.001
33.438
0.9122
Gaussian noise
0.002
31.609
0.8580
45
38.421
0.6952
Rotation
90
36.699
0.8705
(0.03,1)
33.870
0.9472
Brightness
(.06,0.98)
34.630
0.9068
Contrast

Anubhav Kumar was born in Bulandshahr, Uttar

Pradesh, India on March 30, 1986. He received his
B.Tech degree in electronics and communication
engineering at Uttar Pradesh Technical University and
M.Tech degree in electronics and communication
engineering (specialization in image processing) at
Uttar Pradesh Technical University. He is a member of
IEEE, IACSIT and many international societies. He
has also many years research experiences in signal &
image processing and microwave engineering.

IV. CONCLUSION
This paper gives a Nobel watermarking algorithm for color
images based on discrete wavelet transform. The quality of
the watermarked picture has a fine and enhanced nature of
Peak signal to noise ratio. The proposed strategy is robust to
loud, geometric and physical attacks. The experiment results
are better as far as NC values of the extracted watermarks and
PSNR of the watermarked image. At last the proposed
algorithm is simple and has better in secure correspondence

Anuradha received her B.Tech degree in electronics and communication

engineering at Uttar Pradesh Technical University and now she is pursuing
her M.Tech in VLSI design at Rajasthan Technical University.

306